Sunlight Drives Away the Shadows: The Value of Open Source

November 10, 2014 by Dickie Whitaker

In its 20-plus years, catastrophe risk modeling had developed nearly insurmountable barriers to alternative views of risk. Except for the largest companies, insurers and reinsurers have had little choice but to accept proprietary models, which had become very sophisticated but were not transparent to users. Executive SummaryForty insurers, reinsurers and brokers have signed onto the open source Oasis Modelling Framework since its launch in January 2014. Here, Project Director Dickie Whitaker explains how open source brings transparency and choice to the once dark art of catastrophe modeling, removing cost barriers and allowing users to make their own improvements to cat models instead of hoping that the vendor will do so.

Executive Summary

Forty insurers, reinsurers and brokers have signed onto the open source Oasis Modelling Framework since its launch in January 2014. Here, Project Director Dickie Whitaker explains how open source brings transparency and choice to the once dark art of catastrophe modeling, removing cost barriers and allowing users to make their own improvements to cat models instead of hoping that the vendor will do so.

A lack of transparency meant that insurers and reinsurers did not have a full picture of how decisions made by the modelers affected the outputs of the models. These choices affected fundamental model components: the selection of the large catalog of possible events used, the associated perils included, the assessment of damage and the calculation of financial impact, including on business interruption.

Although such choices are necessary to build a manageable system, the assumptions used inevitably result in a certain degree of uncertainty about the results, especially because they can compound each other. Insurers and reinsurers must reflect this uncertainty in their risk capital allocation, and if they suffer higher than expected losses, it can damage their reputation with investors and regulators. Many companies, therefore, have been running several models in an effort to get a more certain view of major catastrophe events such as U.S. hurricane and California earthquake.